Automated Thrombin Generation Assay for Rivaroxaban, Apixaban, and Edoxaban Measurements

Background: The thrombin generation assay (TG) is a promising approach to measure the degree of anticoagulation in patients treated with direct oral anticoagulants (DOAC). A strong association with plasma drug concentrations would be a meaningful argument for the potential use to monitor DOAC. Objectives: We aimed to study the correlation of TG with rivaroxaban, apixaban, and edoxaban drug concentrations in a large, prospective multicenter cross-sectional study. Methods: Five-hundred and fifty-nine patients were included in nine tertiary hospitals. The Technothrombin® TG was conducted in addition to an anti-Xa assay; LC-MS/MS was performed as the reference standard. Results: Correlation (rs) between thrombin generation measurements and drug concentrations was −0.72 for peak thrombin generation (95% confidence interval, CI, −0.77, −0.66), −0.55 for area under the curve (AUC; 95% CI −0.61, −0.48), and 0.80 for lag time (95% CI 0.75, 0.84). In contrast, rs was 0.96 with results of the anti-Xa activity (95% CI 0.95–0.97). Sensitivity with regard to the clinically relevant cut-off value of 50 μgL−1 was 49% in case of peak thrombin generation (95% CI, 44, 55), 29% in case of AUC (95% CI, 24, 34), and 64% in case of lag time (95% CI, 58, 69). Sensitivity of the anti-Xa assay was 95% (95% CI, 92, 97). Conclusions: The correlation of thrombin generation measurements with DOAC drug concentrations was weak, and clinically relevant drug levels were not predicted correctly. Our results do not support an application of TG in the monitoring of DOAC.


INTRODUCTION
Direct oral anticoagulants (DOAC), including the anti-Xa inhibitors rivaroxaban, apixaban, and edoxaban, have simplified and improved health care in many patients worldwide (1). However, whether or not drug monitoring would improve patient outcomes remains unclear. Vitamin K antagonists (VKA) and unfractionated heparin have been successfully applied for decades, and treatment without monitoring would not have been conceivable (2). Low molecular weight heparins (LMWH), which share some similarities in the mechanism of action with anti-Xa inhibitors, are applied using weight-adapted dosages without the need for regular monitoring (3). Indeed, various arguments for the monitoring of DOAC appear (4). First, an association between drug exposure and bleeding events was observed in some studies (5) and between trough concentrations and thromboembolic events such as strokes in other studies (6,7). Second, several clinical factors (e.g., renal failure, liver failure, age > 75 years) and several drug interactions (e.g., P-glycoprotein and CYP3A4-modulating substances) significantly affect drug concentrations (8)(9)(10)(11)(12). Third, a number of observational studies observed a relevant inter-individual variation of the drug concentration (13)(14)(15). Other authors even summarized previous data to suggest an optimal risk-benefit range pointing to a potential target range based on drug concentrations (16). However, the optimal laboratory assay to be used for monitoring of DOAC is to be determined.
Different laboratory tests are being used to measure DOAC drug concentrations in special clinical situations such as major bleeding, urgent surgery, or suspected accumulation (17,18). The anti-Xa inhibitors rivaroxaban, apixaban, and edoxaban can be determined using chromogenic anti-Xa assays (19). These tests, which are successfully used to determine unfractionated and low molecular weight heparins, were further developed and successfully studied in various evaluation studies (15,20). Routine coagulation tests such as prothrombin time are not sensitive to detect clinically relevant concentrations of anti-Xa inhibitors (21). The thrombin inhibitor dabigatran can be captured using the thrombin time, an extremely sensitive routine coagulation test available in many institutions (18). Diluted thrombin times have been developed to measure dabigatran drug concentrations (22,23). Besides, the ecarin clotting time can be utilized to estimate dabigatran drug concentrations (19,22).
The thrombin generation assay (TG) is regarded as a promising approach for monitoring DOAC anticoagulant activity (24). It can record the amount and kinetics of thrombin generated at a particular point in time (25). Thrombin is the key enzyme of the coagulation cascade, which is generated following factor X activation. The generation of thrombin controls, among other functions, the amount of fibrin generated and thus triggers a stable clot (26). The TG measures the cleavage of a fluorogenic substrate over time, resulting in a characteristic curve describing the thrombin generation activity. The parameters of the TG are retrieved from this curve: the maximum concentration of thrombin (thrombin peak height), the total amount of thrombin generation (area under the curve; AUC), and the time to initiation of the exponential phase (lag time). Following these considerations, one might hypothesize that TG is an elegant, functional assay to be used to monitor DOAC, including the anti-Xa inhibitors rivaroxaban, apixaban, and edoxaban. A strong correlation of TG measures with anticoagulant drug concentrations would be a strong argument for a potential application in monitoring DOAC.

Aims
We aimed to study the association of state-of-the-art thrombin generation measurements with rivaroxaban, apixaban, and edoxaban drug concentrations, and compare the results with the performance of an anti-Xa assay in a prospective multicenter cross-sectional study conducted in routine clinical practice.

Design, Setting, and Population
This is a multicenter cross-sectional study conducted between 2018 and 2019 (20). Patients treated with rivaroxaban, apixaban, or edoxaban in routine clinical practice were included in nine specialized hemostasis laboratories affiliated to tertiary hospitals in Switzerland. Inclusion criteria were (a) age above 18 years, (b) use of rivaroxaban, apixaban, and edoxaban, (c) DOAC drug level requested by the treating physician, and (d) signed general informed consent, if requested by the local authorities. Exclusion criteria were (a) refused general informed consent, (b) use of heparin, (c) preanalytical issues, (d) intake of more than one DOAC, and (e) insufficient sample material. To cover the full range of drug levels observed in clinical practice, we collected samples regardless of the time of last drug intake. The study design is illustrated in Figure 1 (CONSORT flow diagram). Ultra-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS) was used as reference (gold) standard because it is regarded as the most accurate technique to measure DOAC drug concentrations (27). The study was approved by the appropriate ethical committees and all participating institutions. The study was conducted in accordance with the declaration of Helsinki.

Data Collection and Handling of Samples
Protocols were implemented at all participating institutions to ensure adequate pre-analytic conditions (28). Blood samples were drawn in plastic tubes containing 1 mL trisodium citrate (0.106 mol L −1 ) per mL of blood. Samples were centrifuged using an established scheme (28), and aliquots were immediately snapfrozen at −20 • C and stored at −80 • C until shipment. Samples were sent on dry ice to the central laboratory and delivered within 3-4 h. Samples were kept frozen until the determination of laboratory tests without any freeze-thaw cycles. All laboratory test results were exported automatically to avoid any typing errors. Coded clinical and laboratory data were stored in a secured RedCAP database. The coding list was stored separately with access by the principal investigator only. The following data were collected: age, sex, and drug.

Thrombin Generation Measurements
An automated thrombin generation assay with high precision characteristics was selected for this study. The Ceveron R TGA RC High (Technoclone, Vienna, Austria) was used on a Ceveron R t100 analyzer (Technoclone, Vienna, Austria), and the manufacturer's instructions were strictly followed. The responsiveness of the Ceveron R TGA RC High trigger reagent for the measurement of direct oral anticoagulants was confirmed in a previous study using spiked samples (29). All lyophilized reagents and calibrators were brought to room temperature, reconstituted with distilled water, and gently mixed. A 4point calibration curve was generated using the Ceveron R TGA CAL set (calibrated against the Thrombin Reference Preparation of the WHO). Before each test run, internal quality control was done using the Ceveron R TGA CONT H and L vials (lyophilized human plasma with increased or decreased thrombin generation). Between-run impression was determined by 19 runs with commercially available frozen normal human plasma (Technoclone, Vienna, Austria). The samples were thawed for 15 min in a water bath at 37 • C and analyzed immediately. To patient's plasma (40 µL), Ceveron R TGA BUF (Tris-Hepes-NaCl buffer), Ceveron R TGA RC High trigger reagent (high concentration of phospholipid micelles containing recombinant human tissue factor in Tris-Hepes-NaCl buffer), FIGURE 1 | Flow of the patients. A prospective, multicentre cross-sectional study was conducted to study the association of a state-of-the-art thrombin generation assay with rivaroxaban, apixaban, and edoxaban drug concentrations in contrast to the association obtained with an anti-Xa assay.

Determination of the Anti-Xa Activity
The TECHNOCHROM R anti-Xa assay was used for determining the anti-Xa activity, using rivaroxaban, apixaban, and edoxaban calibrators (Technoclone, Vienna, Austria).

Determination of the LC-MS/MS
Rivaroxaban, apixaban, edoxaban, and edoxaban M4 were quantified by LC-MS/MS as previously described (20). Briefly, protein precipitation and analyte extraction were performed by adding to the plasma acetonitrile:water 1:1 (v/v), extraction buffer (MassTox TDM Series A, Chromsystems, Gräfelfing, Germany), and precipitation reagent (MassTox TDM Series, Chromsystems, Gräfelfing, Germany) containing the isotope labeled internal standards (13C6 rivaroxaban, 13CD3 apixaban, 13CD2 edoxaban) The samples were vortexed and then centrifuged at 14000 rcf and 20 • C for 4 min. The supernatant was diluted with water: methanol 8:2 (v/v) and stored at 10 • C until analysis. Calibrators and QCs were prepared in pooled plasma. Then, the extracted samples were analyzed by reversed-phase chromatography on a triple quadrupole mass spectrometer (Xevo TQ-S, Waters, Milford, USA) coupled to a UPLC Acquity I-Class system (Waters, Milford, USA). Edoxaban M4 concentration was summed up with edoxaban concentration for further analysis.

Statistical Analysis
The demographic characteristics were described using proportions and percentages and a median and interquartile range (IQR) for the continuous variables. The accuracy of the TG assay and the anti-Xa assay was determined by calculating Spearman's correlation coefficient in relation to the plasma concentration as measured by LC-MS/MS (overall and per drug). A correlation coefficient of r s ≥0.9 was considered as accurate (alternative hypothesis). The Deming regression was used to describe the linear relationship, and a Bland-Altman plot was created to observe a potential bias over the spectrum of measurements (done for the anti-Xa activity only) (30). The systematic differences between the assay are analyzed by calculating the mean difference and the SD to compute 95% limits of agreement for every level of measurements (average difference ± 1.96 standard deviation of the difference) (31). To assess the diagnostic accuracy of the TG and the anti-Xa assay in terms of clinical utility, we determined their sensitivity and specificity with regard to the clinically relevant drug level 50 µg/L, used as a cut-off below which most invasive procedures allowed. A sensitivity of at least 90% and a specificity of more than 80% were regarded as adequate. All statistical analyses were performed using RStudio (1.3. 1093-1); figures were created using Prism 8 (GraphPad Software, Inc., La Jolla, CA, USA).

Patient Characteristics
Nine-hundred and thirty-two patients were included in this prospective multicenter cross-sectional study; the CONSORT flow diagram is given in Figure 1. From this population, 36 patients were excluded because of heparin use, five patients due to pre-analytical issues, two patients because of using more than one DOAC, and 330 patients due to insufficient sample material. Eventually, samples of 559 patients were used for the current analysis. Of those, 300 patients used rivaroxaban, 220 apixaban, and 39 edoxaban. The median age was 75 years (IQR, 65-82 years) and 41.5% of the patients were female. Details are given in Table 1.

Association Between Thrombin Generation Measurements and Drug Concentrations
The association between thrombin generation measurements and drug concentrations are illustrated in Figures 2B-D

DISCUSSION
This study reports 559 patients treated with rivaroxaban, apixaban, or edoxaban in routine clinical practice, which were studied in a cross-sectional study analyzing thrombin generation assay and anti-Xa activity in comparison to LC-MS/MS. The correlation of thrombin generation measurements with drug concentrations was weak, and clinically relevant drug levels were not predicted correctly. In contrast, the correlation of anti-Xa measurements with drug concentrations was high, and relevant drug levels were predicted correctly.
Thrombin  (42). Similar results were also observed in a randomized, placebo-controlled cross-over study including patients taking rivaroxaban (43). Pfrepper et al. included 380 samples from patients taking apixaban, dabigatran, edoxaban, or rivaroxaban and correlated anti-Xa measurements with TG parameters (44). Similar to these studies, we observed changes in TG measures in patients treated with rivaroxaban, apixaban, and edoxaban. However, correlation of TG parameters with drug concentrations was weak to moderate only.
In contrast to previous studies, we included a large number of patients taking rivaroxaban, apixaban, or edoxaban in clinical practice. LC-MS/MS was conducted as a reference standard in all patient samples. Besides, the sensitivity and specificity of TG measurements were calculated against established cut-offs levels. Anti-Xa activity was measured in parallel to compare the results with an assay used routinely in most tertiary hospitals. In addition, patients were recruited from nine tertiary hospitals in Switzerland, which increases the external validity and makes the samples more representative for the general patient population. Therefore, selection bias is considered to be unlikely (45). However, several limitations must be noted. First and most important, we did not record clinical outcomes in terms of bleedings and thromboembolic events. As long as these events are rare, a study like this would need a much larger patient population and a very long observation period. Second, the number of patients is limited in certain subgroups such as lower concentrations of edoxaban. Even though there is no indication that the effects observed might be different in these subgroups, we cannot fully exclude such an effect. Third, our results were obtained with one specific TG device and reagent and the results might be different with other analyzers. However, this is a device of the latest generation with a high precision and there is no hint that this assay could perform worse than others. Fourth, thrombin generation measurements were conducted using frozen citrated samples, and we cannot entirely exclude that the results might be different in whole blood or fresh samples. However, all pre-analytical requirements were strictly addressed, and we do not believe that this might have entirely changed the results.
Following our data, we have to reject the hypothesis that TG measurements adequately reflect DOAC drug concentrations. The association of thrombin generation measurements with drug concentrations was weak, and clinically relevant drug levels were not predicted correctly. We consider our data to be robust because they were obtained in the most accurate study conducted so far. Therefore, our results represent a clear argument against a potential application of TG in monitoring DOAC. One might argue that TG unfolds its utility for monitoring DOAC by assessing the biological efficacy rather than the drug concentration. Several central arguments can be raised supporting this proposition: First, TG reflects the mechanism of thrombus formation more than any other laboratory test (discussed above). Secondly, TG measurements vary among patients with similar drug levels, potentially reflecting different degrees of baseline pro-thrombotic tendency (4). Thirdly, anticoagulant treatment at a given dosage is not effective in some clinical settings representing high-risk pro-thrombotic situations [e.g., DOAC in patients with antiphospholipid antibody syndrome (46); vitamin K-antagonists in cancer patients (47)]. However, this claim must be supported with data demonstrating a relevant association between TG measurements and clinical outcomes in terms of thromboembolic or bleeding events in anticoagulated patients. To the best of our knowledge, these data are not (yet) available. In contrast, the correlation between anti-Xa activity and drug concentrations was strong in our study as well as in others, and an association between anti-Xa activity and clinical outcomes have been shown in various clinical settings (5)(6)(7)(48)(49)(50).

CONCLUSION
We report results of a large prospective study including patients treated with rivaroxaban, apixaban, or edoxaban in routine clinical practice. The correlation of thrombin generation measurements with DOAC drug concentrations was weak, and clinically relevant drug levels were not predicted correctly. Our results do not support an application of TG in the monitoring of DOAC.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Kantonale Ethikkommission Bern. The patients/participants provided their written general informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
TM analyzed the data and wrote the manuscript. J-DS, AM, LA, PF, WW, AS, LG, BG, CB, GM, and TS collected data, contributed to study design, protocol, and preparation of the manuscript. GM conducted the LC-MS/MS measurements. MN designed the study, wrote the protocol, collected and analyzed the data, and wrote the manuscript. All authors contributed to the article and approved the submitted version.